We propose to develop a suite of computational tools for the prediction of tertiary structures of alpha-helical transmembrane (TM) proteins. The core of the-project is to use experimental data-constrained Monte Carlo (MC) simulation to predict the 3D packing of TM helices. To maximally ensure that the MC simulation converges to the global minimum state of the energy suface, we propose to adopt the popular Wang-Landau (WL) algorithm in our implementaion of the MC simulation. The key feature of WL is the elimination of thermally activated energy barriers to possible conformation changes. Our initial folding prediction will be based on existing energy functions such as CHARMM. A new set of energy functions will be developed based on chemical/physical principles and known structures to facilitate more efficient folding calculation. To overcome the problem of searching the enormous conformational space for finding an optimally folded structure, novel MC sampling techniques specifically-tailored for TM proteins will be developed, one of which is to use stable hairpin structures formed by sequentially-neighboring helix pairs as building blocks for the assembly of helical bundles. Also, we propose to develop a systematic approach to generate a small number of distance, orientation, and geometric shape constraints through nuclear magnetic resonance (NMR) and other experiments such as paramagnetic relaxation enhancement, and to apply these experimental data to constrain the search space in the MC simulation. In addition, we propose to develop a threading-based prediction method for membrane proteins, which relies on solved structural homologues or analogs in the Protein Data Bank. Such predictions should generally provide good predictions of the topological arrangements of the TM helices, which could then be used as the starting structures of the MC simulations. The research proposed here, if successful, will provide a never before available capability for protein structural prediction to clinical researchers working on TM proteins, which are the targets for about 2/3 of the contemporary medical drugs. In addition, some of the proposed test systems, such as the amyloid precursor protein (APP) and the beta secretase protein (BACE1) are membrane proteins implicated in the etiology of Alzheimer's disease. The predicted structural information for these proteins could provide a basis for understanding and controlling this disease.